Do teens really pick between AI help with homework and AI help with feelings — or does one quietly turn into the other?

This explores whether teens use chatbots for emotional support as well as schoolwork. The collection's short answer is that the question sets up a choice that may not exist. The evidence points to teens doing both, and to the emotional use often growing out of the practical one. Two surveys show each side. Pew found that 54% of U.S. teens ages 13-17 use chatbots for schoolwork and find them helpful, while 59% think AI cheating is common at their school How do U.S. teens actually use chatbots for schoolwork?. Common Sense Media found that nearly three in four teens have used AI companions, and about a third say they have picked an AI over a person for a serious conversation How many teens are substituting AI companions for human conversation?. One caveat matters here: the Common Sense figure is self-reported preference. It does not measure whether that substitution made teens lonelier or better off.

The more surprising finding is about how emotional use starts. An 18-month study of people using ChatGPT, Claude and Gemini found that emotional reliance usually wasn't the reason they began. It built up slowly through ordinary practical use and turned into routines. Those routines were then disrupted by model updates, news coverage about AI harms, and changes in people's lives How does emotional chatbot use develop from practical use?. That study wasn't limited to teens, but the pattern suggests the homework tool and the confidant can be the same chat window at different moments. The support role also keeps changing; it doesn't settle into a fixed state.

Why would a teen bring something serious to a chatbot? One explanation is that the chatbot doesn't judge. People disclose more when nobody is there to react, and much of the benefit comes from putting their own thoughts into words, not from the bot understanding them Do chatbots help people disclose more intimate secrets?. Chatbots also trigger the human habit of reciprocity. When a bot shares 'feelings' consistently, users open up more in return Do chatbots trigger human reciprocity norms around self-disclosure?. For a teenager, a space without social stakes that seems to open up in return is a strong draw.

The risk is that feeling supported and being helped well are different things. Patients report a real bond with therapeutic chatbots. That bond can coexist with the bot reinforcing unhealthy thinking, or with soothing that blunts emotions that were signaling something Do therapeutic chatbot bond scores hide deeper safety problems?. In tests on health scenarios, models also failed to notice when users were ambivalent or resistant to change, which is often the state of someone who reaches out quietly Why can't chatbots detect when users are ambivalent about change?. A simulated classroom of student agents suggests that how a counseling chatbot actually replies shapes stress, self-reliance and AI dependence over weeks, and that these effects spread to peers How do different counselor styles shape student stress and AI dependence?. It is a simulation, not data from real teens, but it shows why tone matters at scale.

One more pattern is worth knowing. Among adults, people under 30 use chatbots the most and are also the most pessimistic that AI will harm them Why do young adults use chatbots most yet trust them least?. Heavy use and trust don't go together. Young people may lean on these tools while staying wary of them. The collection does not yet include a study that follows teens' emotional chatbot use over time and measures its effect on well-being. That gap matters, because short studies tend to overstate how sticky these relationships are Do chatbot relationships lose their appeal as novelty wears off?.